Examples that match the current platform
These guides are now aligned with the current API To MCP flow: the visual UI Wizard, the manager MCP for AI agents, OAuth 2.0 support, workflow tools, REST, and GraphQL.
How to read these examples
- Each guide starts with the best auth model for that API, not just a list of endpoints.
- Each guide covers both creation paths: the visual builder and the manager MCP.
- Examples focus on production choices: public MCP, internal MCP, or per-user OAuth.
- Workflow tools are included only when they add value beyond a single endpoint call.
Two production-ready creation flows
Visual UI Wizard
Open the builder, choose REST or GraphQL, select the correct auth type, define tools, test them, and deploy.
AI Agent via Manager MCP
Connect the official manager MCP, point your AI agent at the API documentation, and let it create, update, test, and deploy the MCP server from chat.
https://mcp.apitomcp.app/Example library
GitHub
REST or GraphQL, bearer token or per-user OAuth, plus workflow summaries for engineering teams.
YouTube
Public video discovery with API keys or per-user OAuth for creator and channel operations.
Stripe
Private billing and subscription tools for internal teams with stored credentials.
SendGrid
Email operations, contacts, template lookup, and transactional workflows.
Discord
Bot-powered moderation and community operations with safer internal deployment patterns.
Weather
A simple public MCP server example for open APIs and first-time deployments.
X / Twitter
Read-only monitoring with bearer auth or per-user posting with OAuth Authorization Code.
Choosing the right auth model
Open or API-key public data
Best for public APIs, weather, datasets, and low-risk discovery tools that end users can connect to directly.
Stored business credentials
Best for internal tools where one owner-managed API credential should power the live MCP server.
OAuth Authorization Code
Best when every employee or end user should connect their own Google, Meta, CRM, or SaaS account.
Workflow tools
Best when the agent needs one clean tool that chains multiple API steps and returns a simpler result.
Start from the UI or from an agent
If you want the fastest path, use the builder. If you want AI to assemble the server for you, connect the manager MCP and use a prompt like this:
Read the API documentation for this platform and create an MCP server with useful read tools first.
Choose the correct authentication model, add clean tool descriptions, shape the outputs for AI clients,
test the server, then return the deployed MCP URL and any setup steps I still need to complete.